The bacterial strains and plasmids used in this work are indicated in Table 1 . E. coli strains were cultured in Luria-Bertani (LB) medium at 37°C. C. violaceum strains were cultured in LB medium or M9 minimal medium supplemented with 0.1% casein hydrolysate (M9CH) at 37°C (Batista et al., 2019). The cultures were supplemented with kanamycin (50 μg/mL), tetracycline (10 µg/mL), or ampicillin (100 μg/mL), when necessary. Iron deficiency was obtained by the addition of 2,2’-dipyridyl (DP) (Sigma) to the medium, while iron sufficiency was achieved by supplementation with FeSO₄ (Sigma), hemin (Hm) (Sigma), or hemoglobin (Hb) (Sigma).

Null-mutant strains were generated by a previously established allelic exchange mutagenesis protocol (da Silva Neto et al., 2012; Batista et al., 2019; Santos et al., 2020). In-frame null-deletion mutants derived from the wild-type C. violaceum ATCC 12472 strain, with the exception of the ΔcbaCEBAΔchuPRSTUV strain that was obtained using the ΔcbaCEBA mutant as background (Batista et al., 2019). The insertion mutants for the non-ribosomal peptide synthetase (NRPS) genes cbaF and vbaF were obtained by a protocol based on a single recombination event (Batista et al., 2019). For genetic complementation, the chuP, chuR, chuS, chuTUV, and chuPRSTUV genes were amplified by PCR, cloned into the low-copy-number plasmid pMR20, and transferred to the mutant strains by conjugation. The primers used for cloning, sequencing, and mutant confirmation are listed in Supplementary Table 1 .

To achieve iron-limited conditions in M9CH for C. violaceum, we determined the minimal inhibitory concentration (MIC) of DP in this medium, as previously performed in LB medium (Batista et al., 2019). Wild-type C. violaceum overnight cultures were diluted to an optical density at 600 nm (OD₆₀₀) of 0.01 in M9CH, without or with DP (100 µM, 112.5 µM, 125 µM, 132.5 µM, and 150 µM), and grown under agitation (250 rpm) at 37°C. The MIC of 132.5 µM DP for the WT strain was established based on the turbidity of the cultures after 24 h cultivation.

Growth curves were determined in M9CH without or with Hm. Overnight cultures of C. violaceum strains were diluted in 5 mL of M9CH to an OD₆₀₀ of 0.02. Then, a new dilution (1:2) was performed to achieve an OD₆₀₀ 0.01 and the required concentrations of Hm in 200 µL final M9CH in 96-well plates. The plates were incubated at 37°C under moderate orbital agitation in SpectraMax i3 MiniMax Imaging Cytometer (Molecular Devices). The measurements of OD₆₀₀ were recorded every 15 minutes over 24 hours. The experiment was performed in three biological replicates.

The ability of C. violaceum to use Hm and Hb as iron sources was assessed using a nutrition assay (Balhesteros et al., 2017) with some modifications. C. violaceum overnight cultures in M9CH were diluted to an OD₆₀₀ of 1.0 in M9CH. Then, 25 µL of each dilution were embedded in 25 mL of iron-depleted M9CH 0.8% agar (containing 50 µM, 100 µM, 125 µM or 150 µM of DP). Paper discs were added onto the plate surface, and 10 µL aliquots of 100 µM Hm, 20 mM NaOH, 150 µM Hb, and 100 mM NaCl were applied to individual discs. After incubation for 16 h at 37°C, we inspected for growth halos that developed around the discs. The growth area was quantified using the Image J software and normalized by subtracting the disc areas. The experiment was performed in three biological replicates.

The toxic concentrations of Hm for C. violaceum were assessed by cell viability. Overnight cultures were diluted to an OD₆₀₀ of 0.01 in M9CH without or with Hm (30 µM, 600 µM, and 2000 µM), and grown under agitation (250 rpm) for 24 h at 37°C. Serial dilution in phosphate-buffered saline (PBS) was performed, and 10 µL were spotted onto M9CH plates. Hemin toxicity was determined based on the colony-forming units (CFU) displayed by the strains after incubation for 24 h at 37°C. These experiments were performed in three biological replicates.

The hemolytic activity was assessed in 5% (v/v) sheep-blood Mueller-Hinton agar plates. Five microliters of C. violaceum M9CH overnight cultures were spotted onto the plate. The hemolytic activity was detected by the lighter halos that developed due to erythrocyte lysis after incubation for 7 days at 37°C. The area of activity was quantified using the Image J software and normalized by subtracting the bacterial growth area. The experiment was performed in three biological replicates.

Siderophores were detected by chrome azurol S (CAS) plate assay in modified peptone-sucrose agar (PSA-CAS) plates (Batista et al., 2019; Santos et al., 2020). Ten microliters of C. violaceum overnight cultures in M9CH were spotted onto the plate surface, and the siderophores were detected by the orange halos that developed after incubation for 24 hours at 37°C. The area of the halos was quantified using the Image J software and normalized by subtracting the bacterial growth area. The experiment was performed in three biological replicates.

The upstream regions of genes of interest were amplified by PCR with specific primers ( Supplementary Table 1 ) and cloned into the pGEM-T easy plasmid (Promega). After digestion with proper restriction enzymes ( Supplementary Table 1 ), the inserts were subcloned into the pRKlacZ290 vector to generate transcriptional fusions to the lacZ gene. C. violaceum cultures harboring the reporter plasmids were grown until an OD₆₀₀ of 0.6 – 0.8 in M9CH, and either untreated or treated with 100 µM Hm or 100 µM FeSO₄ for 2 h. For all expression assays, the M9CH medium was used as the iron-limited condition because we previously found that the expression of an iron-regulated gene was similarly high in M9CH or M9CH with DP (Santos et al., 2020). Bacterial cells were assayed for β-galactosidase activity as previously described (Santos et al., 2020). The experiment was performed in three biological replicates.

The C. violaceum wild-type strain was grown in M9CH until an OD₆₀₀ of 1.0 – 1.2. Total RNA was extracted using Trizol reagent (Invitrogen) and purified with Direct-zol™ RNA Miniprep Plus (Zymo Research). RT-PCR was performed with the SuperScript III One-Step RT-PCR System with Platinum Taq High Fidelity DNA Polymerase (Invitrogen). One microgram of each RNA sample and specific primers ( Supplementary Table 1 ) that amplify regions from chuP to chuR (439 bp), chuR to chuS (373 bp), and chuS to chuT (662 bp) were used in the reactions. PCRs using conventional Taq DNA polymerase, and the same sets of primers, were performed with genomic DNA (positive control) and RNA (negative control) as templates.

The C. violaceum wild type, ΔchuP, and ΔchuP[chuP] strains were grown in M9CH until midlog growth phase, and the cultures were either untreated or treated with 100 µM Hm or 100 µM FeSO₄ for 2 h. Total RNA was extracted and purified as described above. Two micrograms of total RNA from each sample were converted to cDNA using the High-Capacity cDNA Reverse Transcription kit (Thermo Fisher Scientific). Genomic DNA contamination (for RNA) and reverse transcription efficiency (for cDNA) were checked by conventional PCR with the primers for the rpoH gene ( Supplementary Table 1 ). Quantitative PCR (qPCR) reactions were performed using the PowerUp™ SYBR™ Green Master Mix (Thermo Fisher Scientific), the specific primers ( Supplementary Table 1 ), and 0.5 µL of cDNA. The relative expression was calculated by the 2^-ΔΔCt method (Livak and Schmittgen, 2001). Data from three biological replicates were normalized by an endogenous control (rpoH gene) and a reference condition (WT in M9CH 100 µM Hm). The treatment with Hm was used as a control based on the β-galactosidase assays that indicated an intermediate expression of the chu operon under this condition.

The coding region of chuP was amplified by PCR ( Supplementary Table 1 ) and cloned into the pET-15b plasmid ( Table 1 ). After induction in E. coli BL21(DE3) with 1 mM Isopropyl β-D-1-thiogalactopyranoside (IPTG) for 2 h at 37°C, the His-ChuP protein was purified from the soluble extract by affinity chromatography in a Ni-NTA Superflow column (Qiagen). The elution fractions were evaluated using 18% SDS-PAGE. The aliquots containing the purified His-ChuP were concentrated using a VivaSpin 6 column (Sartorius), and desalted by gel filtration in PD-10 column (GE Healthcare) in storage buffer (100 mM NaH₂PO₄, 600 mM NaCl, 20% glycerol, pH 8) (Puri and O’Brian, 2006). The concentration of the His-ChuP protein was determined by measurement of OD at 280 nm and using its extinction coefficient calculated by the Protparam Tool (ExPASy) (http://web.expasy.org/protparam).

The ability of the recombinant His-ChuP protein to interact with Hm was evaluated by spectrophotometry (Puri and O’Brian, 2006; Amarelle et al., 2016). The reactions were performed in interaction buffer (50 mM Na₂H₂PO₄, 300 mM NaCl, 10% glycerol, pH 8) without (reference cuvette) or with 10 µM of His-ChuP (sample cuvette). Aliquots of Hm (0 to 30 µM) were added to both cuvettes. After incubation for 5 minutes at 25°C in the dark, the absorbance between the wavelengths 300 and 600 nm was measured with 10 nm increments on a SpectraMax i3 MiniMax Imaging Cytometer. The binding of ChuP to Hm was determined by the change in absorbance at 413 nm fit to one-site binding model non-linear regression on Graph Pad Prism 7.

DNA sequences upstream of chuP, chuR, and CV_2599 were amplified by PCR using the primers listed in Supplementary Table 1 . The DNA fragments were radiolabeled and used for interaction with His-ChuP following a previously described protocol (da Silva Neto et al., 2009; Previato-Mello et al., 2017), with the modification of adding the CV_2599 promoter fragment (negative control) in the same reaction.

Virulence assays were performed in a mouse intraperitoneal (i.p.) model of C. violaceum infection as previously established (Previato-Mello et al., 2017; Batista et al., 2019). Bacterial strains were diluted to an OD₆₀₀ of 0.01 and cultured in 5 mL LB for 20 h at 37°C. A dose of 10⁶ CFU in PBS was injected into 6-week-old female BALB/c mice, and the animals were monitored for 7 days post-infection. To assess the bacterial burden in the liver and spleen, mice were infected as above and euthanized 20 h or 96 h post-infection (h.p.i.). The organs were aseptically collected, homogenized in PBS, and the dilutions were plated for CFU counting. Mice were obtained and maintained at the Animal Facilities of Ribeirão Preto Medical School (FMRP-USP). The assays were performed according to the Ethical Principles in Animal Research adopted by the National Council for the Control of Animal Experimentation (CONCEA). The animal ethics protocol 146/2019 was approved by the Local Ethics Animal Committee (CEUA) of FMRP-USP.

Data collected were employed for statistical analysis in GraphPad Prism version 7. For the column graphs, the normality test was performed using Shapiro-Wilk’s test. Statistically significant p values and the tests that were performed are indicated in the figure legends.

In silico analysis of the C. violaceum ATCC 12472 genome sequence revealed a gene cluster with six genes (CV_RS19275-280-285-290-295-300) that resembles an operon encoding a putative heme utilization system. These genes, here named chuPRSTUV, are annotated as a HemP/HmuP family regulator (ChuP), a TonB-dependent receptor (ChuR), a hemin degrading factor (ChuS), and an ABC-transport system (ChuTUV) ( Figure 1A ). To evaluate if the chuPRSTUV genes are organized into an operon, we performed RT-PCR reactions using RNA from the WT strain grown in M9CH and a set of primers that amplify regions between chuPR, chuRS, and chuST genes ( Figures 1A, B ). After reverse transcription and amplification, bands with the expected sizes were detected for the three tested primer combinations, confirming that the chuPRSTUV genes are indeed co-transcribed ( Figure 1B ).

Our inspection of the promoter region of chuP revealed a putative Fur binding site sequence (ATGATAATGGTTATCATT) that resembles Fur boxes found in other bacteria (Sarvan et al., 2018). To investigate whether the chu operon is regulated by iron and Fur, we cloned the promoter region of chuP into a lacZ reporter plasmid. The WT and Δfur strains harboring the pchuP-lacZ fusion were used to assess the chuP promoter activity by β-galactosidase assay in M9CH medium, which was previously reported as an iron-limited condition (Santos et al., 2020) and under iron sufficiency (M9CH supplemented with Hm or FeSO₄) ( Figure 1C ). The promoter activity was higher under iron-limited (M9CH) than iron-replete conditions in the WT strain. The reduction in activity was higher with FeSO₄ than with Hm supplementation. In the Δfur mutant, the promoter was highly active regardless of iron levels. Moreover, the activity was twofold higher than that detected for the WT strain in M9CH, suggesting total promoter de-repression in the absence of Fur ( Figure 1C ). Altogether, these results demonstrate that the chuPRSTUV genes comprise a Fur-repressed operon that is expressed under iron limitation.

To characterize the role of the chuPRSTUV operon in C. violaceum, we generated null-mutant strains deleted for single genes (ΔchuP, ΔchuR, and ΔchuS) or multiple genes (ΔchuTUV and ΔchuPRSTUV) of the chu operon, and their respective complemented strains. We also obtained a mutant strain lacking both the chu operon and the cbaCEBA genes (ΔcbaCEBAΔchuPRSTUV). The CbaCEBA enzymes are involved in the synthesis of 2’3-DHB, the precursor of catecholate-type siderophores in C. violaceum (Batista et al., 2019). All mutants showed regular fitness, as assessed by growth curves in M9CH and M9CH plus heme and by cell viability in LB ( Supplementary Figure 1 ).

To test the involvement of the C. violaceum chu genes in heme and hemoglobin utilization, we developed a nutrition assay providing 100 µM Hm or 150 µM Hb as alternative iron sources in M9CH medium chelated for iron with different DP concentrations ( Supplementary Figure 2 ). We chose 125 µM DP to compare all strains ( Figure 2 ) because it was the best condition to visualize the growth halos in the WT strain ( Supplementary Figure 2 ). Under these conditions (125 µM DP), the WT and the ΔchuS strains formed Hm and Hb-stimulated growth halos. All the other mutant strains of the chuPRSTUV operon lost the ability to grow when Hm and Hb were provided as iron sources ( Figure 2 ). For the ΔchuR strain, a very weak growth stimulus could still be detected only in the presence of heme ( Figure 2B ). Genetic complementation of the mutant strains fully restored the growth in Hm and Hb under deficiency with 125 µM DP ( Figure 2 ). The ΔcbaCEBA mutant that does not synthesize siderophores showed no growth halos at 125 µM DP ( Figure 2 ), but its growth was clearly stimulated by Hm and Hb at 50 µM DP ( Supplementary Figure 2 ). This is consistent with previous results indicating that the growth of a ΔcbaCEBA mutant is strongly impaired under DP-imposed iron limitation (Batista et al., 2019). Taken together, these results demonstrate that the chuPRSTUV operon encodes a heme uptake system (ChuRTUV) that is also involved in hemoglobin utilization. Moreover, the weak growth detected for the ΔchuR mutant with Hm but not with Hb suggests that C. violaceum has additional mechanisms in the outer membrane for heme uptake but relies specifically on ChuR for heme uptake from hemoglobin.

Considering that the ΔchuS strain showed no altered phenotype for Hm and Hb utilization ( Figure 2 ), we evaluated its role on cell viability under heme excess ( Supplementary Figure 3 ). However, growth defects were not observed for the WT, ΔchuS, and all mutant strains even at a high Hm concentration of 2 mM, indicating that the chu operon has no role during our heme excess conditions. Interestingly, deletion of the chuPRSTUV operon in the ΔcbaCEBA mutant strain improved the small colony size phenotype ( Supplementary Figure 3 ), previously described for this strain (Batista et al., 2019). We also tested the hemolytic activity of the chu mutants on sheep-blood agar ( Supplementary Figure 4 ). The strains ΔchuR (increased and intense halo) and ΔcbaCEBA (intense halo) showed altered hemolytic activity when compared to that of the WT and the other mutant strains. Although the meaning of these findings is unclear, we speculate that the increased hemolytic activity in these strains is a compensatory mechanism to deal with iron/heme scarcity.

To verify whether the chu operon affects the production/release of siderophores in C. violaceum, we tested the chu mutants on PSA-CAS plates for siderophore detection as orange halos ( Figure 3 ) as previously described (Batista et al., 2019). The ΔchuP and ΔchuPRSTUV mutants showed increased siderophore halos, while the ΔchuR, ΔchuS, and ΔchuTUV mutants had siderophore halos similar to that of the WT strain ( Figures 3A, B ). The ΔcbaCEBA strain showed no siderophore halo, as previously demonstrated (Batista et al., 2019), as well as the ΔcbaCEBAΔchuPRSTUV strain ( Figures 3A, B ). After complementation, the siderophore halos were restored to WT levels in the ΔchuP[chuP] strain. For the strains ΔchuPRSTUV[chuPRSTUV] (almost absence of halo) and ΔcbaCEBA[cbaCEBA] (increased halo), the siderophore phenotypes reverted further on that observed in the WT strain ( Figures 3A, B ), perhaps owing to overexpression of the genes into the plasmid. These data indicate that the small regulatory protein ChuP controls the siderophore levels in C. violaceum.

C. violaceum produces the catecholate-type siderophores chromobactin and viobactin employing the NRPS enzymes CbaF and VbaF, respectively (Batista et al., 2019). We combined mutation in chuP with mutations in cbaF or vbaF to understand which siderophore contributes to the increased siderophore halos in ΔchuP. Individual deletion of cbaF or vbaF genes in the WT had no effect on siderophore halos ( Figures 3C, D ), as previously reported (Batista et al., 2019). When these genes were deleted in the ΔchuP mutant background, a decrease in siderophore halos was observed in both cases. However, the halos were similar to that of the WT strain only when vbaF was deleted ( Figures 3C, D ), demonstrating a prominent role of viobactin on the increased siderophore halos of ΔchuP. Altogether, these results suggest that ChuP controls the synthesis and/or uptake of the siderophore viobactin in C. violaceum.

Our data indicate that mutation of chuP in C. violaceum abolished heme utilization ( Figure 2 ) and altered the levels of the siderophore viobactin ( Figure 3 ). We employed different methodologies to elucidate how ChuP regulates these processes ( Figure 4 ). First, we tested whether ChuP is a heme-binding protein. We purified the recombinant protein His-ChuP and performed a heme-binding assay ( Figure 4A ). After incubation with increasing Hm concentrations, a Soret peak at 413 nm was detected, indicating the formation of a ChuP-heme complex ( Figure 4A ). The differential absorption spectroscopy at 413 was used to fit a single binding model and determined that ChuP binds heme with a k_d = 18.36 ± 4.66 µM ( Figure 4A , insert). Considering that HemP/HmuP proteins have been described as transcriptional activators (Escamilla-Hernandez and O’Brian, 2012; Sato et al., 2017), we tested whether the C. violaceum ChuP regulates and binds into the intergenic regions upstream of chuP (promoter of the chu operon) and chuR ( Figures 4B, C ). The WT and ΔchuP strains harboring these constructs (pchuP-lacZ or pchuR-lacZ) were assessed by β-galactosidase assay under different iron levels. The pchuP-lacZ promoter fusion was highly active under iron deficiency (M9CH) with a gradual decrease upon Hm and FeSO₄ supplementation ( Figure 4B ), as previously observed in the WT strain ( Figure 1C ). However, the same activity pattern was detected in the ΔchuP mutant strain, indicating that ChuP does not seem to regulate the promoter of the chu operon ( Figure 4B ). The pchuR-lacZ fusion had no promoter activity regardless of the strain or condition, indicating the absence of a promoter upstream of chuR. Therefore, this fusion is not useful to verify the effect of ChuP on chuR expression. Consistent with the β-galactosidase assays, our EMSA assays indicated that ChuP does not bind to the probes containing only the promoter of the chu operon or containing the entire region from chuP to chuR ( Figure 4C ). Altogether, these results demonstrate that ChuP does not regulate the promoter of the chu operon nor act as a DNA binding protein.

In E. meliloti, the HmuP protein activates the expression of the TBDR ShmR at a post-transcriptional level, probably by acting on a sequence HPRE (HmuP-responsive element). The HPRE sequences were predicted upstream of genes encoding heme TBDRs in many bacteria, including the C. violaceum chuR (sequence CCCGCAAGCCAGCCGACAGCCAGCCAGCG, -26 nt from the ATG start codon) (Amarelle et al., 2019). In addition to chuR, we found an HPRE sequence upstream of vbuA (sequence GCCAGCCAGACGACGCCGCCG, -49 nt from the ATG start codon), a TBDR gene located far from the chu operon ( Figures 4D, E ), raising the possibility that ChuP is a post-transcriptional regulator in C. violaceum. To verify this hypothesis, we performed RT-qPCR for chuR, vbuA, and vbaF genes with RNA harvested from the WT, ΔchuP, and ΔchuP[chuP] under different iron levels ( Figures 4D–F ). The expression of the three genes in the WT strain was high under iron-depleted and low under iron-sufficient conditions when compared to the control condition (WT grown in M9CH 100 µM Hm), as expected for genes related to iron acquisition ( Figures 4D–F ). Consistent with our phenotypic results and the presence of HPRE elements, the expression of chuR and vbuA was decreased in the ΔchuP strain regardless of the iron levels ( Figures 4D, E ), indicating that ChuP is a positive regulator required for the maximum expression of chuR and vbuA under iron limitation. Complementation of ΔchuP restored the expression of chuR and vbuA to the levels found in the WT strain ( Figures 4D, E ). No differences in vbaF expression were detected between the WT and ΔchuP strains in any of the tested conditions ( Figure 4F ), indicating that ChuP does not control the expression of VbaF, the NRPS for viobactin synthesis. Therefore, the decreased expression of chuR and vbuA in ΔchuP explains the inability of this mutant strain to use Hm and Hb ( Figure 2 ) (via ChuR) and its increased siderophore halos ( Figure 3 ) (inability to uptake viobactin via VbuA). Indeed, a ΔvbuA mutant showed large siderophore halos (Batista et al., 2019) as those found in ΔchuP. Altogether, these results demonstrate that ChuP integrates the acquisition of heme and siderophore by acting as a heme-binding post-transcriptional regulator of the TBDR genes chuR and vbuA.

To assess the role of the heme utilization system ChuPRSTUV during C. violaceum infection, we performed mice virulence assays ( Figure 5 ). The animals were i.p. injected with a dose of 10⁶ bacterial cells and analyzed for survival during seven days post-infection ( Figures 5A, B ). The five null-mutant strains of the ChuPRSTUV system showed barely or no virulence attenuation compared to the C. violaceum WT strain ( Figure 5A ). Previously, we determined that abrogating siderophore production in C. violaceum by deletion of the cbaCEBA genes causes moderate attenuation in virulence (Batista et al., 2019). Therefore, we checked whether heme and siderophores cooperate for virulence. Indeed, the ΔcbaCEBA strain showed an intermediate virulence attenuation, as expected, while a more expressive virulence attenuation was observed for the ΔcbaCEBAΔchuPRSTUV strain ( Figure 5A ). Complementation of the latter strain with the chuPRSTUV operon reverted its virulence attenuation phenotype to the pattern observed for the ΔcbaCEBA mutant ( Figure 5B ).

We evaluated the bacterial burden in the liver and spleen, two organs colonized during C. violaceum infection that are involved in host heme recycling. Interestingly, the ΔcbaCEBAΔchuPRSTUV mutant displayed the same CFU counting as the WT strain at 20 hours post-infection in both organs ( Figures 5C, D ). However, the bacterial burden was reduced (in the liver) and eliminated (in the spleen) at 96 hours post-infection ( Figures 5C, D ). These results indicate that the absence of siderophores and heme uptake does not impair initial colonization but impairs the bacterial maintenance in later infection stages. Altogether, these results indicate an interplay between the iron-acquisition strategies based on siderophore and heme during C. violaceum infection. In our acute infection model, the requirement of heme uptake for virulence becomes evident in the absence of siderophores.